Case Study · Built & Run by DM AI Labs

DM Devotional — an AI content pipeline, not a one-off demo.

A devotional YouTube channel — mythology films, 3D-animated songs, photo-realistic music videos — scripted, generated, edited and published by one person using an AI production pipeline. This page shows the actual videos, the actual process, and the actual numbers.

269VIDEOS SINCE 10 JULY
82KVIEWS · SINCE LAUNCH
5,143WATCH HOURS · SINCE LAUNCH
Why it exists

Devotional content is expensive to make well — traditionally.

A single mythology film or classical-style bhajan video normally means a studio, actors, VFX, and weeks of production. DM Devotional runs the same output through an AI pipeline instead — one person, a script, a set of character sheets, and a chain of AI tools that handles generation, editing and publishing.

Every video below is live on the channel right now — not a private demo reel.
Seven videos, five playlists, one pipeline

Stories, 3D-animated songs, photo-realistic songs, and mantras.

Same process end to end — only the visual style, playlist and prompt direction change. Click any card to watch it on YouTube.

Veerbhadra: Lord Shiva's Wrath
►
Stories

Veerbhadra: Lord Shiva's Wrath

Full Shiv Puran Animated Story · Complete Film · 21:47

The full Veerbhadra saga in one film — Sati's devotion, Daksha's fatal pride, the birth of Veerbhadra, and the divine war at Daksha's Yagya. Five shorter parts were produced first and stitched into this complete version.

683views in 4 days
41.5watch hours
+463views vs. this channel's usual
Ek Lota Jal
►
Songs · 3D animation

Ek Lota Jal

Bhole Baba Devotional Song · 3D Shiv Story · 3:09

A short, fully 3D-animated devotional song built around a single simple ritual — pouring water over a Shivling. Character and prop models were generated once and reused across every shot for visual consistency.

68%average retention
2:09avg. watch time of 3:09
1.4watch hours
Radha Ki Tamanna Hai
►
Songs · Photo-realistic

Radha Ki Tamanna Hai

Krishna Devotional Bhajan · Official Video · 6:54

A cinematic, photo-realistic Radha-Krishna piece — lighting, fabric and skin rendered close to live-action rather than illustrated or animated. The channel's strongest single performer to date.

1.9Kviews
40.0watch hours
+1.7Kviews vs. this channel's usual
Karna's Sacrifice
►
Stories · Mythology

Karna's Sacrifice

Mahabharat Animated Story · Mythology Stories playlist · 6:01

Karna's defining act — giving away his armor and earrings knowing it would cost him the war. One of the pipeline's strongest mythology performers outside the Veerbhadra film.

658views
15.7watch hours
+518views vs. this channel's usual
Bappa Morya
►
Songs · Ganesha

Bappa Morya

Ganpati Visarjan Devotional Song · Ganesha playlist · 2:33

A short Ganpati Visarjan song timed to the immersion-season search spike — built for when people are actively searching for visarjan content, not just browsing.

116views
1.6watch hours
+76views vs. this channel's usual
Sanjeevani
►
Stories · Mythology

Sanjeevani

Hanuman Lifts the Mountain · Mythology Stories playlist · 3:04

The Ramayana's Sanjeevani episode — Hanuman carrying the entire mountain to save Lakshman. Two-thirds of this video's traffic came from YouTube search rather than suggested or browse.

152views
2.2watch hours
67%traffic from YouTube search
Ganpati Baba
►
Mantra · Ganesha

Ganpati Baba

Obstacle-Removal Mantra · Ganesha playlist · 3:11

One of the simplest formats in the pipeline — a single mantra, a steady visual, a clean loopable chant. Built to rank on search rather than lean on suggested traffic.

160views
2.4watch hours
36days since publish
Browse the channel

Every playlist on DM Devotional.

How it's made

Eight steps from a blank page to a published video.

The same pipeline runs for every video on the channel, regardless of format.

01

Concept & script

Claude and ChatGPT turn a theme — a deity, a festival, an episode from the Puranas — into a structured script: narration beats for a story, verses for a song. Mythological accuracy gets checked here, before anything is generated.

02

Character sheets

Before a single frame is generated, the recurring characters — Shiva, Krishna, Ganesha — get a locked character sheet: face, palette, proportions, attire. Every later prompt references it, so the same deity looks like themself in shot 40 as in shot 1.

03

Video prompts

The script is broken into shot-by-shot prompts — camera angle, lighting, motion, art style (photo-realistic vs. 3D-animated) — written with Claude and ChatGPT, then fed into Google Flow (Veo) for generation.

04

Generation & stitching

Individual AI-generated clips come back from Google Flow and get assembled, timed to music and polished — cuts, transitions, pacing — in YouTube Create and Edits.

05

Music

A bhajan or background track is sourced and matched to the visuals through YT Music.

06

Titles, thumbnail & SEO

Canva builds the click-worthy thumbnail. Claude and ChatGPT draft the title, description, hashtags and keyword set — optimized against what VidIQ shows is actually being searched.

07

SEO check before publish

VidIQ scores the draft (title, tags, description) against competing content and search volume before anything goes live — not after.

08

Publish & iterate

YouTube Studio handles upload and scheduling. Real performance — views, watch time, retention, traffic source — is tracked from there and feeds directly into the next script.

The stack

Nine tools, one person running all of them.

No production team — this is what one person operating an AI pipeline actually uses, end to end.

Claude

Scripts, character sheets, shot-by-shot video prompts, titles, descriptions, hashtags

ChatGPT

Second-pass scripting and prompt refinement

Google Flow (Veo)

AI video generation from prompts — both photo-realistic and 3D-animated styles

YouTube Create

Stitching, cuts and assembly of generated clips

Edits

Fine timing, transitions and polish

YT Music

Background scores and bhajan tracks

Canva

Thumbnail design

VidIQ

Keyword research, SEO scoring, competing-content analysis

YouTube Studio

Publishing, scheduling, analytics and retention tracking

The numbers

What this pipeline actually produces.

Channel-wide performance since the channel launched on 10 July 2026 — pulled directly from YouTube Studio.

82KViews
5,143Watch time (hours)
269Videos published since 10 July
Read honestly: subscriber conversion is low relative to views — typical of browse- and suggested-video-driven devotional content, where people watch and move on rather than subscribe. We lead with views and watch hours here because they're what prove the pipeline actually produces watchable content at volume — subscriber count is a slower, separate number we're not going to dress up.
Proof, not just claims

The dashboards behind the numbers above.

Screenshots straight from YouTube Studio and VidIQ — the same 28-day window, cross-checked across two independent tools. Captured in August 2026, so these are a point-in-time snapshot and read lower than the running totals above.

Watch time trend, last 28 days
YouTube Studio · Channel · Last 28 days

Watch time trend, last 28 days

Channel-wide watch-time curve pulled straight from YouTube Studio — no subscriber numbers, just the trend.

Top content this period
YouTube Studio · Content · Last 28 days

Top content this period

The actual top-performing videos, ranked by real audience retention and views — not a curated highlight reel.

Top video: watch time over 28 days
YouTube Studio · Per-video · Last 28 days

Top video: watch time over 28 days

A single video's watch-time chart on an explicit 28-day window, including average view duration.

Top video: reach & impressions
YouTube Studio · Per-video · Last 28 days

Top video: reach & impressions

Impressions, click-through rate, views and unique viewers for the same video, last 28 days.

From impression to watch time
YouTube Studio · Per-video · 28 days

From impression to watch time

The full funnel — impressions, CTR, views from impressions, average duration, and watch time it produced.

How viewers find this video
YouTube Studio · Per-video · Last 28 days

How viewers find this video

Traffic-source breakdown — mostly suggested videos and playlists, which is typical for this content type.

Cross-checked in VidIQ
VidIQ · Channel stats overlay · Last 28 days

Cross-checked in VidIQ

The same watch-time and views figures, verified independently in a second analytics tool.

Same numbers, second tool
VidIQ · AI Coach overview

Same numbers, second tool

VidIQ's own channel overview and growth chart, corroborating what YouTube Studio shows above.

Want an AI content pipeline like this for your brand?

Same method behind everything we build at DM AI Labs — script it, hand the repeatable parts to AI, keep a human on the judgment calls. Book a free 30-minute call and we'll look at where it fits your content.

Book your free 30-min call